anti-angiogenic activity of the purine analog 6-thioguanine

Leukemia (2002) 16, 1490–1499 2002 Nature Publishing Group All rights reserved 0887-6924/02 $25.00

www.nature.com/leu

Anti-angiogenic activity of the purine analog 6-thioguanineM Presta1, M Belleri1, A Vacca2 and D Ribatti3

1Unit of General Pathology and Immunology, Department of Biomedical Sciences and Biotechnology, School of Medicine, University ofBrescia, Brescia, Italy; 2Department of Biomedical Sciences and Human Oncology, University of Bari, Bari, Italy; and 3Department of HumanAnatomy and Histology, University of Bari, Bari, Italy

The antimetabolite 6-thioguanine (6-TG) is utilized in the man-agement of acute myelogenous leukemia (AML). Angiogenesisis a possible therapeutic target in hematologic tumors. Thus,we addressed the possibility that 6-TG may also act as an anti-angiogenic molecule. 6-TG inhibited endothelial cell prolifer-ation triggered by fibroblast growth factor-2 (FGF2) and vascu-lar endothelial growth factor (VEGF) and delayed the repair ofa mechanically wounded endothelial cell monolayer. Also, 6-TG inhibited sprouting within fibrin gel, morphogenesis onMatrigel, and collagen gel invasion by endothelial cells. 2-Ami-nopurine was ineffective. In vivo, 6-TG inhibited basal, VEGF-induced, and FGF2-induced vascularization in the chickembryo chorioallantoic membrane and prevented neovascu-larization triggered by leukemia LIK cells or their conditionedmedium. Finally, bone marrow vascularization in AML patientswas decreased to control values in the early remission phaseand persisted unvaried after 8–12 months of maintenance ther-apy with 6-TG. Thus, 6-TG inhibits different steps of the angiog-enesis process in vitro and exerts a potent anti-angiogenicactivity in vivo. Its anti-angiogenic activity, together with itsantimetabolite activity towards tumor cells, may contribute toits action during maintenance therapy in AML. These resultssuggest a new rationale for the use of purine analogs in themanagement of AML.Leukemia (2002) 16, 1490–1499. doi:10.1038/sj.leu.2402646Keywords: angiogenesis; purine analog; FGF2; VEGF

Introduction

In the adult, the proliferation rate of endothelial cells is verylow compared to many other cell types in the body. Uncon-trolled endothelial cell proliferation is observed in tumor neo-vascularization, angioproliferative diseases like Kaposi’s sar-coma, and angiogenesis-dependent diseases like rheumatoidarthritis, psoriasis, and a number of eye diseases.1 Variousangiogenesis inhibitors have been developed so far, their effi-cacy has been evaluated in different in vitro and in vivoassays,2 and their clinical evaluation in cancer patients is inprogress (for further information about angiogenesis inhibitorsin clinical trials see the NCI web site:http://cancertrials.nci.nih.gov). Recently, the hypothesis thatanti-angiogenic compounds can be used in combination withcytotoxic drugs for tumor therapy has been advanced (see Ref.3 and references therein). Also, chemotherapeutic agents haveshown anti-angiogenic properties in vitro and in vivo,4,5 lead-ing to the concept of anti-angiogenic scheduling of chemo-therapy.5,6

Purine analogs were developed in the early 1950s as anti-neoplastic chemotherapeutic agents.7 These antimetabolites

Correspondence: M Presta, Unit of General Pathology and Immu-nology, Department of Biomedical Sciences and Biotechnology,School of Medicine, University of Brescia, via Valsabbina 19, 25123Brescia, Italy; Fax: +39-0303701157Received 2 April 2002; accepted 14 May 2002

inhibit de novo purine synthesis and purine interconversionreactions and their metabolites can be incorporated intonucleic acids.7 6-Thioguanine (6-TG) and 6-methylmercapto-purine riboside (6-MMPR) also alter membrane glycoproteinsynthesis.8 Purine analogs can act as protein kinase inhibitors,6-MMPR showing a high potency and selectivity for nervegrowth factor-activated protein kinase N.9 2-Aminopurine (2-AP) inhibits proto-oncogene and interferon gene transcrip-tion.10 Combination chemotherapy regimens for the manage-ment of solid tumors have been proposed in which purineanalogs are administered in association with cytotoxicdrugs.11,12 At present, the purine analog 6-TG is used in themanagement of acute myelogenous leukemia (AML) both inremission induction and in maintenance therapy.7

An increasing body of evidence points to a role for bonemarrow angiogenesis in hematologic tumors.13–15 Forinstance, bone marrow vascularization is increased in patientswith AML.15–17 Also, AML cells produce angiogenesis factors,including fibroblast growth factor-2 (FGF2) and vascular endo-thelial growth factor (VEGF),15 whose levels are an inde-pendent predictor of outcome.18

Recently, we demonstrated that the purine analog 6-MMPRmodulates the angiogenic activity of FGF2 in vitro and affectsblood vessel formation in vivo.19 Also, continuous systemicadministration of 6-mercaptopurine ribose phosphate inhibitsangiogenesis in the rabbit cornea.5 In contrast, 6-methylmer-captopurine, 2-AP, and adenine are devoid of anti-angiogenicactivity,19 thus indicating that subtle structural differences maydetermine the ability of purine analogs to affect neovasculariz-ation. On this basis, we addressed the possibility that 6-TGmay act as an anti-angiogenic molecule, this activity contribu-ting to its efficacy in AML therapy.6-TG was evaluated for the capacity to affect various steps

of the angiogenesis process (ie cell proliferation, motility,endothelial cell sprouting, collagen invasion, and formationof capillary-like structures) induced by FGF2 and/or VEGF incultured endothelial cells of different origin. The in vitroobservations were compared to the effect of 6-TG on in vivoneovascularization in the chick embryo chorioallantoic mem-brane (CAM) under basal conditions or during neovasculariz-ation induced by FGF2 or VEGF or by human leukemia LIKcells grafted on to the CAM. Finally, we evaluated bone mar-row vascularization in AML patients given maintenance ther-apy with 6-TG. The results demonstrate that 6-TG inhibits dif-ferent steps of the angiogenesis process in vitro and exerts apotent anti-angiogenic activity in the CAM. Its anti-angiogeniccapacity, together with its antimetabolite activity, may con-tribute to its action during maintenance therapy in AMLpatients.

6-Thioguanine and angiogenesisM Presta et al

1491Materials and methods

Reagents

6-TG and 2-AP were purchased from Sigma Chemical (StLouis, MO, USA). Human recombinant FGF2 was expressedand purified from E. coli cell extract by heparin–Sepharoseaffinity chromatography as described.20 The 165 amino acidisoform of VEGF (VEGF165) was from Calbiochem (San Diego,CA, USA).

Cell cultures

Fetal bovine aortic endothelial GM 7373 cells were obtainedfrom the NIGMS Human Genetic Mutant Cell Repository(Institute for Medical Research, Camden, NJ, USA) and grownin Eagle’s minimal essential medium containing 10% fetal calfserum (FCS) (Integro, Zaandam, The Netherlands), vitamins,essential and non-essential amino acids. Immortalized Balb/cmurine aortic endothelial (MAE) cells (clone 22106) andmurine brain microvascular endothelial (MBE) cells (clone10027) were obtained from R Auerbach (University of Wis-consin, Madison, WI, USA) and were grown in Dulbecco’smodified minimal essential medium (DMEM; Gibco, LifeTechnologies, Rockville, MD, USA) added with 10% FCS.FGF2-T-MAE cells represent a highly tumorigenic subclone ofFGF2-transfected MAE cells.21 They express high levels ofFGF2 and form highly vascularized tumors in nude mice.21

FGF2-T-MAE cells were grown in DMEM supplemented with4 mM glutamine (Gibco) and 10% FCS. Bovine aortic endo-thelial (BAE) cells (provided by A Vecchi, Istituto Mario Negri,Milan, Italy) were cultured in MEM-Eagle’s medium sup-plemented with 10% FCS, 2% essential amino acids and 2%vitamins. Cultures were used between the 6th and the 10thcell passage. LIK cells were obtained from the American TypeCulture Collection (Rockville, MD, USA) and grown in RPMI-1640 medium (Gibco) supplemented with 10% heat inacti-vated FCS and 1% glutamine.

Cell proliferation assays

GM 7373 cells were seeded at 70 000 cells/cm2. After over-night incubation, cells were incubated for 24 h in freshmedium containing 0.4% FCS and 10 ng/ml FGF2 in the pres-ence of increasing concentrations of purine analogs. At theend of the incubation, cells were trypsinized and counted ina Burker chamber. Under these experimental conditions, cul-tures incubated in 0.4% FCS or 0.4% FCS plus FGF2 undergo0.1–0.2 and 0.7–0.8 cell population doublings, respectively.22

MAE cells were seeded at 25 000 cells/cm2. After overnightincubation, cells were starved for 48 h in fresh medium con-taining 0.5% FCS. Then, cells were treated with 30 ng/mlFGF2 in the absence or presence of purine analogs. After 18h, cells were incubated with 3H-thymidine (1 �Ci/ml) for afurther 6 h and radioactivity incorporated into the TCA-insol-uble material was measured. FGF2-T-MAE cells (75 000/cm2)were incubated with complete medium in the absence or inthe presence of purine analogs. After 24 h, cells were tryp-sinized and counted in a Burker chamber.

Leukemia

Wounding of endothelial cell monolayers

Endothelial cells were allowed to reach confluence. Then,wounds were created in the cell monolayer with a 1.0-mmwide rubber policeman. Cultured medium and detached cellswere removed and monolayers were incubated in freshmedium added with 10% FCS and the purine analog undertest.20 The number of cells migrating into the wound was mea-sured by computerized image analysis 36–48 h after wound-ing.

Preparation of three-dimensional gels

Fibrin gels were prepared as described,23 with minor modifi-cations. Briefly, MBE or FGF2-T-MAE cell aggregates, pre-pared on agarose-coated plates exactly as described,24 wereseeded on to fibrin-coated 24-well plates. Immediately afterseeding, 500 �l of calcium-free medium containing fibrinogen

Figure 1 Effect of 6-TG on endothelial cell proliferation. (a) GM7373 cells were seeded at 70 000 cells/cm2. After overnight incu-bation, cells were incubated in fresh medium containing 0.4% FCSand 10 ng/ml of FGF2 (circles) or 30 ng/ml VEGF165 (triangles) in thepresence of increasing concentrations of 6-TG (open symbols) or 2-AP (closed symbols). After 24 h, cells were trypsinized and counted(a). In (b) FGF2-treated GM 7373 cells were incubated for 3 days withvehicle (open squares), 30 �M 6-TG (open circles) or 30 �M 2-AP(closed circles). Then, culture medium was changed to fresh mediumwith no addition (arrow) and cells were counted at the indicatedtimes. (c) MAE cells were seeded at 25 000 cells/cm2. After starvation,cells were treated with 30 ng/ml FGF2 in the absence or presence of6-TG (open circles) or 2-AP (closed circles). After 18 h, cells wereincubated for a further 6 h with 3H-thymidine and DNA incorporatedradioactivity was measured (arrow: radioactivity incorporated inFGF2-untreated cells). (d) FGF2-T-MAE cells were seeded at75 000/cm2. After 24 h (T0, gray bar), cells were incubated with com-plete medium in the absence (hatched bar) or in the presence of 6-TG (open bars) or 2-AP (black bars). After 24 h, cells were trypsinizedand counted. All the experiments were repeated three times with simi-lar results. **Statistically different from untreated cell cultures (P �0.01 or better).


1492

Leukemia

(2.5 mg/ml) and thrombin (250 mU/ml) were added to eachwell and allowed to gel for 5 min at 37°C. Then, 500 �l ofculture medium were added on the top of the gel. Culturemedium was renewed every 48 h. When present FGF2 andpurine analogs were added to both fibrin gel and cell culturemedium at the indicated concentrations. In all the experi-ments the fibrinolytic inhibitor trasylol was added to the geland to the culture medium at 200 KIU/ml to prevent the dis-solution of the substrate.23 Formation of radially growing cellsprouts was observed during the next 2–5 days.Matrigel (10 mg/ml; Becton Dickinson, Milan, Italy) was

used to coat 48-well plates at 4°C (150 �l/well). After gel-ification at 37°C, FGF2-T-MAE cells were seeded on to Matri-gel-coated dishes at 75 000 cells/cm2 in the absence or in thepresence of purine analogs. Newly formed endothelial cell‘cords’ and ‘tubes’ were photographed using an invertedphase contrast photomicroscope.For the preparation of three-dimensional gels of reconsti-

tuted collagen fibrils, 7 volumes of 1.5 mg/ml rat tail tendontype I collagen (Boehringer Mannheim Italia, Milan, Italy) dis-solved in 0.1% acetic acid were mixed on ice with 2 volumesof 5× concentrated medium containing NaHCO3 and 1 vol-ume of 250 mM Hepes. The pH of the mixture was balancedby alkaline solution containing 1.0 N NaOH and 22 mg/mlNaHCO3. The mixture was allowed to solidify in 24-wellplates (0.4 ml/well) at 37°C. Then, BAE cells were seeded onthe top of collagen gel and allowed to reach confluence. Cellcultures were then treated with FGF2 plus VEGF165 (both at30 ng/ml) in the absence or in the presence of purine analogs.After 24 h, cells were photographed using an inverted phasecontrast photomicroscope and endothelial cells invading thegel, in a plane of focus beneath the cell monolayer surface,were quantified by computerized analysis of the digitalizedimages.

Figure 2 Effect of 6-TG on the repair of wounded endothelial cell monolayers. Confluent monolayers of GM 7373 cells were wounded witha 1.0-mm wide rubber policeman. Then, cultures were incubated in fresh medium in the absence (a) or in the presence of 100 �M 2-AP (b)or 6-TG (c). After 2 days, cells were fixed and photographed. Dotted lines indicate the width of the wound at T0. In (d) MAE cell monolayerswere wounded and incubated with complete medium in the absence (hatched bar) or in the presence of the indicated concentrations of 6-TG(open bars) or 2-AP (black bars). After 36 h, the number of cells invading the wound per ×40 microscopic field was measured by computerizedimage analysis. Cytochalasin D at 0.5 �g/ml was used as a control (T0, gray bar). Each point is the mean ± s.d. of three determinations.**Statistically different from untreated cell cultures (P � 0.01 or better).

Chorioallantoic membrane (CAM) assay

Fertilized White Leghorn chicken eggs were incubated underconditions of constant humidity at a temperature of 37°C. Onthe third day, a square window was opened in the shell afterremoval of 2–3 ml of albumen to detach the shell from thedeveloping CAM. The window was sealed with a glass of thesame size and the eggs were returned to the incubator.In a first set of experiments, 10 �l-methylcellulose disks,

containing 5 nmols of 6-TG, 2-AP, or 400 ng of affinity-pur-ified anti-FGF2 antibody (provided by DB Rifkin, New YorkUniversity Medical Center, New York, NY, USA), wereimplanted on the top of chick embryo CAM at day 8. At day12, CAM were processed for light microscopy and the angiog-enic response was assessed by a planimetric method of ‘pointcounting’.25 Briefly, every third section within 60 serial slidesfrom an individual specimen was analyzed by a 144-pointmesh inserted in the eyepiece of a Leitz-Dialux 20 photomic-roscope. Six randomly chosen microscopic fields of each sec-tion were evaluated at ×160 magnification for the total num-ber of the intersection points which were occupied by vesselstransversally cut (diameter ranging from 3 to 10 �m). Meanvalues ± standard deviation for vessel counts were determinedfor each analysis. The vascular density was indicated by thefinal mean number of the occupied intersection points as per-cent of the total number of intersection points. The statisticalsignificance of differences among the mean values of the inter-section points in the experimental series was determined byStudent’s t-test for unpaired data. The planimetric method of‘point counting’ was also utilized for the quantitation offibroblast density within CAM mesoderm.In a second set of experiments, gelatin sponges (Gelfoam

Upjohn Company, Kalamazoo, MI, USA) were cut to a sizeof 1 mm3 and placed on top of the CAM at day 8, under sterile


1493conditions.26 The sponges were then adsorbed, in the absenceor in the presence of 6-TG (5–50 nmols/sponge), with PBS,FGF2, or VEGF165 (both at 1 �g/sponge), 1 �l of LIK cell sus-pension (20 000 cells/sponge), or 3 �l of their five-fold con-centrated conditioned medium prepared as described.27 Atday 12, CAMs were photographed in ovo with a stereomicro-scope equipped with a Camera System MC 63 (Zeiss, Oberk-ochen, Germany). Blood vessels entering the sponges withinthe focal plane of the CAM were counted by two observersin a double-bind fashion at ×50 magnification.28

Serial evaluation of bone marrow neovascularizationin AML patients

Bone marrow biopsies from five patients with AML(males/females, 5/2; median age, 65 years) were studied atdiagnosis, after remission induction therapy (in months 1–3)and during remission maintenance therapy (in months 8–12).Diagnosis and classification (three patients: M2; one patient:M3; one patient; M5) were done according to the criteria ofthe French–American–British (FAB) Cooperative Group.29

Remission induction therapy consisted of cytosine arabinoside(100 mg/m2 daily for 7 days), daunorubicin (45 mg/m2 dailyfor 3 days), and etoposide (75 mg/m2 daily for 7 days). Afterremission, patients were given maintenance therapy with 6-TG (Tabloid brand Thioguanine, 2 mg/kg body weight daily

Figure 3 Effect of 6-TG on in vitro endothelial cell sprouting and morphogenesis. (a–c) MBE cell aggregates were embedded in fibrin gel inthe presence of 30 ng/ml FGF2, incubated with vehicle (a), 2-AP (b) or 6-TG (c), and photographed after 5 days. (d–f) FGF2-T-MAE cell aggregateswere embedded in fibrin gel in the absence (d) or in the presence of 2-AP (e) or 6-TG (f) and photographed after 2 days. (g–i) FGF2-T-MAEcells were seeded on Matrigel in the absence (g) or in the presence of 2-AP (h) or 6-TG (i) and photographed after 2 days. In all the experimentspurine analogs were at 30 �M. Sprouting and morphogenesis were observed in controls and in 2-AP-treated cultures but were fully inhibitedin 6-TG-treated cultures.

Leukemia

or every other day). Control subjects were eight patients withanemia due to iron or vitamin B12 deficiencies (males/females,5/3; median age, 48 years). The study protocol was approvedby the Research Ethics Committee of the University of Bariand informed consent was obtained from all patients. Biopsieswere studied for microvessel density as described.14 Briefly,6-�m deparaffinized and hydrated sections were sequentiallyincubated with a monoclonal (murine, IgG1) antibody to theendothelial cell marker factor VIII-related antigen (FVIII-RA),a biotin-labelled (horse) anti-mouse IgG and streptavidin–per-oxidase conjugate (all from Dako, Glostrup, Denmark), thenred-stained with a 3-amino-9-ethylcarbazole solution andcounterstained with hematoxylin. Microvessels, ie capillariesand small venules, were identified as endothelial cells, eithersingle or clustered in nests or tubes, and clearly separatedfrom one another, and either without or with a lumen (notexceeding 10 �m). Megakaryocytes also stained by FVIII-RAwere easily distinguishable by their size and morphology. Fourto six 250× fields covering each of two sections per biopsy wereexamined within a superimposed 484-point square grid of 12.5× 10−2 mm2. The area occupied by microvessels, or ‘microvesselarea’, was calculated by using the planimetric method of ‘pointcounting’, according to which the area equalled the sum ofpoints (129 �m2 per point) that hit microvessels. The microvesselarea was normalized to the cellular area (grid area minus denseconnective tissue, bone lamellae, fat, necrosis and hemorrhagicareas) by computerized image analysis.


1494

Leukemia

Results

Effect of 6-TG on cultured endothelial cells

To evaluate a possible angiosuppressive activity of 6-TG, thedrug and its analog 2-AP, here used as a negative control,19

were tested for the capacity to inhibit cell proliferation inendothelial GM 7373 cells treated with FGF2 or VEGF. Asshown in Figure 1a, 6-TG inhibits cell proliferation triggeredby the two angiogenic growth factors with a similar potency.Also, 6-TG inhibits FGF2-induced proliferation in MAE cellsand in FGF2-overexpressing endothelial FGF2-T-MAE cells inthe �M range of concentrations (Figure 1c, d). 2-AP wasinstead ineffective in all the endothelial cell types tested. Also,when FGF2-treated GM 7373 cells were incubated for 3 dayswith 30 �M 6-TG and then the culture medium was changedto fresh medium with no addition, the number of 6-TG-treatedcells remained unchanged for the next 5 days (Figure 1b). Itmust be pointed out, however, that cell viability was stillhigher than 95% after 8 days of treatment with 6-TG as evalu-ated by trypan blue staining of trypsinized cells (data notshown). Taken together, these data indicate that 6-TG exertsa potent cytostatic effect on endothelial cells.6-TG also significantly affected the capacity of GM 7373

cells to repair a mechanically wounded cell monolayer. Acomplete repair was observed 2 days after wounding in con-trol cultures (Figure 2a) and in cultures treated with 100 �M2-AP (Figure 2b), whereas a significant reduction in cellmigration occurred in cultures treated with 100 �M 6-TG(Figure 2c). Similarly, 10 �M and 100 �M 6-TG inhibited therepair of mechanically wounded monolayers of MAE cells(Figure 2d).The capacity of 6-TG to affect different steps of the angiog-

enesis process was investigated further by the in vitro sproutformation assay.23 In this assay, endothelial cell aggregates areembedded into the fibrin gel in the presence of angiogenicstimuli and the formation of radially growing endothelialsprouts follows. Accordingly, MBE cell aggregates invade thegel and form solid sprouts after 2–3 days in culture when incu-bated in the presence of 30 ng/ml FGF2. 6-TG (30 �M) fullyprevented sprout formation whereas 2-AP was ineffective(Figure 3a–c).In vitro, the culture of different endothelial cell types on

Matrigel, a laminin-rich gelled basement membrane matrix,results in the formation of vascular tubes, a phenomenonknown as ‘spontaneous angiogenesis’.30 Previous observationsin our laboratory had shown that FGF2-T-MAE cells invade3D fibrin gel and undergo morphogenesis on Matrigel.31 Asshown in Figure 3d-i, 6-TG was able to inhibit fibrin invasionand vascular tube formation on Matrigel by FGF2-T-MAE cellswhen tested at 30 �M. Again, 2-AP was ineffective.Finally, 6-TG suppressed the ability of BAE cells to invade

a 3D collagen gel when stimulated by FGF plus VEGF. Noactivity was instead exerted by 2-AP (Figure 4).In conclusion, 6-TG inhibits the in vitro different steps of the

angiogenesis process including endothelial cell proliferation,migration, sprout formation, invasion, and morphogenesis.

Effect of 6-TG on the vascularization of the chickembryo CAM

On the basis of the in vitro observations, 6-TG was evaluatedfor the capacity to affect the basal growth of new blood vesselsin vivo in the chick embryo CAM.25 The macroscopic vascular

Figure 4 Effect of 6-TG on collagen gel invasion by endothelialcells. BAE cells were seeded on the top of collagen gel and allowedto reach confluence (T0). Cell cultures were then treated with vehicle(a) or FGF2 plus VEGF165 (both at 30 ng/ml) in the absence (b) or inthe presence of 30 �M 6-TG (c) or 2-AP (d). After 24 h, cells werephotographed using an inverted phase contrast photomicroscope in aplane of focus beneath the cell monolayer surface. In (e), cell mono-layers were incubated with FGF2 plus VEGF165 in the absence(hatched bar) or in the presence of the indicated concentrations of 6-TG (open bars) or 2-AP (black bars). After 24 h, endothelial cellsinvading the gel were quantified by computerized analysis of the digi-talized images. Each point is the mean ± SD of three determi-nations. **Statistically different from untreated cell cultures (P � 0.01or better).

pattern of CAMs exposed to 5 nmols 2-AP delivered from day8 to day 12 of development by a methylcellulose disk implantwas indistinguishable from that of control embryos (Figure 5a,b). In contrast, an avascular zone free of blood vessels wasevidenced beneath the implants containing 5 nmols of 6-TG(Figure 5c), similar to that observed in embryos treated withneutralizing anti-FGF2 antibody (not shown, see Ref. 13). His-


1495

Figure 5 Effect of 6-TG on physiological CAM vascularization. (a–c) Macroscopic observations. Five nmols of 6-TG or 2-AP absorbed onmethylcellulose disc were implanted on to the CAM of 8-day embryo and the anti-angiogenic response was evaluated at day 12. Compare theavascular zone corresponding to the site of implant of 6-TG (asterisk in c) with the normal morphology of the vasculature in CAM treated withvehicle (asterisk in a) or with 2-AP (asterisk in b). (d–g) Semithin sections. In (d) control CAM at day 12: the CAM is made up of an chorionicepithelium (c), a capillary meshwork (arrowheads) running under the ectoderm, an intermediate mesenchyme containing large vessels (v) andfibroblasts, and a deep allantoic epithelium (a). A similar vascularization is observed in 2-AP-treated CAM (e). No blood vessels are recognizablebeneath the chorion and in the mesoderm of CAMs treated with 6-TG (f). In (g) CAM treated with anti-FGF2 antibody (400 ng/implant) showingno blood vessels beneath the ectoderm and in the intermediate mesenchyme, loosely arranged fibroblasts. Original magnification: (a–c), ×30;(d–g) ×250.

tological observation of CAM sections at day 12 showed thatthe CAM of control embryos was formed by flat chorion andallantoic epithelia with capillary blood vessels located at thebase of the chorion and by the mesoderm containing largearteries and veins, fibroblasts, and few leukocytes (Figure 5d).Normally developed blood vessels were also detectable in theCAM of 2-AP-treated embryos (Figure 5e). No blood vesselswere instead recognizable beneath the chorion and in themesoderm of CAMs treated with 6-TG or anti-FGF2 antibody(Figure 5f, g). Quantitation of blood vessel density by a plani-metric method of ‘point counting’ confirmed the morphologi-cal observations and point to an anti-angiogenic effect for 6-TG in vivo (Table 1). Interestingly, at variance with anti-FGF2antibody, 6-TG administration did not affect fibroblast density

Table 1 Effect of 6-TG on CAM vascularization

Treatment Microvessel density Fibroblast density

Intersection points Area Intersection points Area(mean ± s.d.) (% of total) (mean ± s.d.) (% of total)

Control (vehicle) 8.0 ± 1.8 5.5 22 ± 4 15.22-AP (5 nmols/implant) 7.0 ± 0.8 4.9 24 ± 3 16.76-TG (5 nmols/implant) 1.0 ± 0.2a 0.7 21 ± 3 12.5Anti-FGF2 Ab (400 ng/implant) 3.0 ± 0.7a 2.0 5 ± 1b 3.6

Methylcellulose discs were implanted on to the CAM of 8-day embryos (20 animals per group). Vascularization and fibroblast density wereassessed at day 12 by a planimetric method of ‘point counting’.aP � 0.001 vs control and 2-AP; b�0.001 vs all the other groups.

Leukemia

in CAM mesoderm (Table 1), ruling out a possible non-spe-cific cytotoxic effect of this compound.In a second set of experiments, 6-TG delivered via a gelatin

sponge implant26 inhibited both basal, FGF2-induced, andVEGF-induced CAM neovascularization with similar doseresponses (Figure 6a). To assess whether 6-TG was also ableto inhibit hematologic tumor-induced angiogenesis, 6-TG wasevaluated for the capacity to affect neovascularization trig-gered by leukemia LIK cells when grafted on to the CAM atday 8 via the gelatin sponge implant. In agreement with pre-vious observations,27 LIK cells implanted on to the CAMinduced a potent angiogenic response that was significantlyreduced by 6-TG given at 25 nmols/implant (Figure 6b). Asimilar inhibitory effect was observed when angiogenesis wasinduced by the conditioned medium of LIK cells (Figure 6b).


1496

Leukemia

Figure 6 Effect of 6-TG on FGF2, VECF, and LIK cell-inducedCAM vascularization. One mm3 gelatin sponges were placed on topof the CAM at day 8 under sterile conditions. The sponges were thenadsorbed with PBS, FGF2, or VEGF165 (both at 1 �g/sponge), 1 �l ofLIK cell suspension (20 000 cells/sponge) or PBS, or 3 �l of their five-fold concentrated conditioned medium (LIK CM) or of control five-foldconcentrated fresh medium (FM) in the absence or in the presence of6-TG (5–50 nmols/sponge). At day 12, blood vessels entering thesponges within the focal plane of the CAM were counted by twoobservers in a double-bind fashion at ×50 magnification. Each pointis the mean ± s.d. of 10 embryos. **Statistically different from 6-TG-untreated embryos (P � 0.01 or better).

Effect of 6-TG on bone marrow vascularization inAML patients

Five AML patients were assessed for bone marrow angiogen-esis at diagnosis. Then, they underwent remission inductiontherapy for 7 days followed by maintenance therapy with 6-TG (2 mg/kg body weight daily or every other day). Bone mar-row angiogenesis was assessed again in the same patients bothin the early (months 1–3) and in the late (months 8–12) phasesof maintenance therapy when all patients were in completeremission. In agreement with previous observations,15–17 AMLpatients were characterized by intense bone marrow angiog-enesis when compared to control patients (Figure 7 and Table2). However, the FVIII-RA-positive microvessel area wasmarkedly decreased in the early remission phase and persistedunvaried after 8 to 12 months of maintenance therapy with6-TG. In addition, changes in microvessel morphology wereobserved in the bone marrow of all AML patients during ther-apy. Indeed, blood vessels at diagnosis were thin, winding,

often without evident lumen, with single spread endothelialcells and small endothelial sprouts (Figure 7a, d). In contrast,vessels were straight, small, and no endothelial sprouts wereevident both in the early (Figure 7b, e) and in the late (Figure7c, f) remission phase.

Discussion

New blood vessel formation is a multi-step process that beginswith the degradation of the basement membrane by activatedendothelial cells that will migrate and proliferate, leading tothe formation of solid endothelial cell sprouts into the stromalspace. Then, vascular loops are formed and capillary tubesdevelop with formation of tight junctions and deposition ofnew basement membrane.32 Each step of this process rep-resents a potential target for the inhibitory action of angiostaticmolecules.2 In the present paper we describe the anti-angiog-enic activity of the purine analog 6-TG. Our results indicatethat 6-TG affects various cell proliferation-independentaspects of the angiogenesis process (ie endothelial cellmotility, sprout formation, collagen gel invasion, andmorphogenesis) at concentrations that are cytostatic but notcytotoxic for endothelial cells. This suggests that 6-TG acts asan angiostatic molecule on multiple targets of the angiogenicprocess. This hypothesis is supported by the observation thatthe anti-angiogenic activity exerted in vivo by this purine ana-log on the physiological blood vessel development of thechick embryo CAM is not the mere consequence of a genericinhibitory action on cell proliferation, as demonstrated by thelack of effect of 6-TG administration on fibroblast density ofthe CAM mesoderm.6-TG is used in the management of AML both in remission

induction and in maintenance therapy.7 Previous observationshave shown that various leukemia cell lines produce andrelease significant amounts of angiogenic growth fac-tors.16,27Accordingly, they trigger an angiogenic response indifferent in vivo animal models, including the CAM assay.27

Here, 6-TG inhibits new blood vessel formation induced byleukemia LIK cells grafted on to the CAM. This appears to bedue to a direct angiosuppressive effect rather than to anindirect activity of the drug on LIK cells. Indeed, 6-TG wasalso able to inhibit neovascularization triggered by the con-ditioned medium of LIK cells or by recombinant FGF2 orVEGF.Purine analogs can act as protein kinase inhibitors9 and

modulate gene expression in different experimental sys-tems,10,11 thus indicating that this class of compounds caninterfere with intracellular signaling and growth factor activi-ties.19,33,34 Moreover, 6-TG and 6-MMPR alter membrane gly-coprotein synthesis.8 Thus, besides the well characterizedeffects of purine analogs as antimetabolic drugs able to inhibitde novo purine synthesis and purine interconversion reac-tions,7 various mechanisms of action may be responsible forthe anti-angiogenic activity of 6-TG. Further experiments arerequired to clarify this point. Also, the chemical structure ofpurine analogs appears to be of importance for their capacityto act as angiogenesis inhibitors. Indeed, 6-TG (present work),6-MMPR,19 and 6-mercaptopurine ribose phosphate5 arepotent angiostatic compounds whereas the structurally related2-AP, 6-methylmercaptopurine, and adenine are ineffective.13

A detailed analysis on a larger number of compounds will berequired to better define the structure–function relationship ofangiostatic purine analogs.Compelling experimental evidence indicates that bone mar-


1497

Figure 7 Bone marrow vascularization in AML patients given 6-TG during maintenance therapy. Factor VIII-RA immunostaining of bonemarrow biopsies from two representative AML patients who were treated with remission induction therapy for 7 days, attained completeremission, and were then given maintenance therapy with 6-TG. Numerous microvessels were detectable at diagnosis in both patients (a, d).In the remission phase (b, month 2; e, month 3) vessels were rare (arrowheads). A very limited vascularization was observed also during themaintenance therapy with 6-TG (c, month 8; f, month 12). Note several strongly stained megakaryocytes as internal positive controls (b, c, eand f). Bar = 60 �m (a–f).

Table 2 Bone marrow vascularization in AML patients

Patients/Status Microvessel area Cellular area(mm2 × 10−2) (mm2 × 10−2)

AML patientsat diagnosis 0.47 ± 0.22 (0.28–1.01)a 11.5 ± 0.8 (9.2–12.3)after complete remission 0.12 ± 0.03 (0.07–0.18) 10.6 ± 0.8 (9.5–12.0)during 6-TG maintenance therapy 0.11 ± 0.02 (0.08–0.15) 10.5 ± 0.8 (9.4–11.6)

Control patients 0.08 ± 0.01 (0.07–0.09) 10.3 ± 0.5 (9.1–10.5)

Bone marrow biopsies from five AML patients were studied for microvessel density at diagnosis, after remission induction therapy (months1–3), and during maintenance therapy with 6-TG (months 8–12). Control subjects (n = 8) were patients with anemia due to iron or vitaminB12 deficiencies. Microvessel area was calculated by using the planimetric method of ‘point counting’ and was normalized to the cellulararea by using a computerized image analysis software. Values are expressed as mean ± s.d. (range).aP � 0.001 vs both remission periods and controls by paired Wilcoxon test.

Leukemia


1498

Leukemia

row angiogenesis plays an important role in hematologictumors.13–15 Here, we have shown that bone marrow vascu-larization is significantly increased in AML patients at diag-nosis in respect to control patients. This is in keeping withprevious findings on bone marrow vascularization in largercohorts of AML patients15–17 and with the observation that theplasma levels of various angiogenic growth factors areincreased in these patients.15,18 Previous observations havealso shown that induction chemotherapy using the TAD proto-col (standard-dose cytarabine, daunorubicin and 6-TG)35

causes a significant decrease of bone marrow vascularizationin AML patients without residual leukemic blast infiltration butnot in patients with residual disease.17 In our patients, bonemarrow vascularization was reduced to control values afterremission induction by cytosine arabinoside, daunorubicinand etoposide treatment, and remained unvaried during main-tenance therapy with 6-TG (up to 12 months after remission).Even though these data do not formally prove that 6-TG actsdirectly on bone marrow endothelium in AML patients, the invitro and in vivo evidence about its anti-angiogenic activity,including its capacity to inhibit CAM neovascularizationinduced by leukemia LIK cells and their conditioned medium,strongly supports the hypothesis that the angiostatic capacityof 6-TG, together with its antimetabolite activity, may contrib-ute to its action in AML therapy.Leukemic cells release endothelial growth factors.16,27,36,37

In turn, activated endothelial cells release cytokines thatstimulate leukemia cell growth.38 Our data suggest that 6-TGmay interrupt this reciprocal loop of stimulation between leu-kemic and endothelial cells by acting on both cell types. Inter-estingly, a similar mechanism of action has been hypothesizedfor arsenic trioxide and vinblastine in the therapy of hematol-ogic tumors.39,40 The combined therapeutic approach tar-geting both tumor cells (by conventional cytotoxic agents) andendothelial cells (by angiogenesis inhibitors) may lead to syn-ergistic antitumor effects.3,5 The anti-angiogenic action ofcytostatic chemotherapy may therefore contribute to the eradi-cation of leukemic cells.In conclusion, our data demonstrate that 6-TG exerts a

potent anti-angiogenic activity in vitro and in vivo thus sug-gesting a new rationale for its use in the management of AMLin remission induction and in maintenance therapy.

Acknowledgements

This work was supported in part by grants from MURST(Centro di Eccellenza ‘IDET’, Cofin 2000, and ex 60%), CNR(Progetto Finalizzato Biotecnologie), and Istituto Superiore diSanita (AIDS Project) to MP, from Associazione Italiana per laRicerca sul Cancro to MP and AV, and from Associazione Ital-iana per la Lotta al Neuroblastoma to DR.

References

1 Folkman J. Angiogenesis in cancer, vascular, rheumatoid andother diseases. Nat Med 1995; 1: 27–31.

2 Keshet E, Ben-Sasson SA. Anticancer drug targets: approachingangiogenesis. J Clin Invest 1999; 104: 1497–1501.

3 Klement G, Baruchel S, Rak J, Man S, Clark K, Hicklin DJ, BohlenP, Kerbel RS. Continuous low-dose therapy with vinblastine andVEGF receptor-2 antibody induces sustained tumor regressionwithout overt toxicity. J Clin Invest 2000; 105: R1–R8.

4 Miller KD, Sweeney CJ, Sledge GW. Redefining the target: chemo-therapeutics as antiangiogenics. J Clin Oncol 2001; 19: 1195–1206.

5 Browder T, Butterfield CE, Kraling BM, Shi B, Marshall B, O’ReillyMS, Folkman J. Antiangiogenic scheduling of chemotherapyimproves efficacy against experimental drug-resistant cancer. Can-cer Res 2000; 60: 1878–1886.

6 Hanahan D, Bergers G, Bergsland E. Less is more, regularly: metro-nomic dosing of cytotoxic drugs can target tumor angiogenesis inmice. J Clin Invest 2000; 105: 1045–1047.

7 Calabresi P, Chabner BA. Chemotherapy of neoplastic diseases.In: Goodman Gilman A, Rall TW, Nies AS, Taylor P (eds). ThePharmacological Basis of Therapeutics. Pergamon Press: NewYork, 1990, pp 1202–1263.

8 Sokoloski JA, Sartorelli AC. Inhibition of mannose incorporationinto glycoproteins and dolichol-linked intermediates of Sarcoma180 cells by 6-methylmercaptopurine ribonucleoside. Int J Cancer1987; 39: 764–768.

9 VolontŁ C, Greene LA. 6-Methylmercaptopurine riboside is apotent and selective inhibitor of nerve growth factor-activated pro-tein kinase N. J Neurochem 1992; 58: 700–708.

10 Zinn K, Keller A, Whittemore L-A, Maniatis T. 2-Aminopurineselectively inhibits the induction of �-interferon, c-fos, and c-mycgene expression. Science 1988; 240: 210–213.

11 O’Dwyer PJ, Hudes GR, Colofiore J, Walczak J, Hoffman J, LaC-reta FP, Comis RL, Martin DS, Ozlos RF. Phase I trial of fluoroura-cil modulation by N-phosphonacetyl-L-aspartate and 6-methyl-mercaptopurine riboside: optimization of 6-methylmercaptopurine riboside dose and schedule through bio-chemical analysis of sequential tumor biopsy specimens. J NatlCancer Inst 1991; 83: 1235–1240.

12 Stolfi RL, Colofiore JR, Nord LD, Koutcher JA, Martin DS. Bio-chemical modulation of tumor cell energy: regression of advancedspontaneous murine breast tumors with a 5-fluoracil-containingdrug combination. Cancer Res 1992; 52: 4074–4081.

13 Perez-Atayade AR, Sallan SE, Tedrow U, Connors S, Allred E, Folk-man J. Spectrum of tumor angiogenesis in the bone marrow ofchildren with acute lymphoblastic leukemia. Am J Pathol 1997;150: 815–821.

14 Vacca A, Ribatti D, Presta M, Minischetti M, Iurlaro M, Ria R,Albini A, Bussolino F, Dammacco F. Bone marrow neovasculariz-ation, plasma cell angiogenic potential and matrix metalloprotein-ase-2 secretion parallel progression of human multiple myeloma.Blood 1999; 93: 3064–3073.

15 Aguayo A, Kantarjian H, Manshouri T, Gidel C, Estey E, ThomasD, Koller C, Estrov Z, O’Brien S, Keating M, Freireich E, Albitar M.Angiogenesis in acute and chronic leukemias and myelodisplasticsyndromes. Blood 2000; 96: 2240–2245.

16 Hussong JW, Rodgers GM, Shami PJ. Evidence of increased angi-ogenesis in patients with acute myeloid leukemia. Blood 2000;95: 309–313.

17 Padro T, Ruiz S, Bieker R, Burger H, Steins M, Kienast J, BuchnerT, Berdel WE, Mesters RM. Increased angiogenesis in the bonemarrow of patients with acute myeloid leukemia. Blood 2000; 95:2637–2644.

18 Aguayo A, Estey E, Kantarjian H, Mansouri T, Gidel C, Keating M,Giles F, Estrov Z, Barlogie B, Albitar M. Cellular vascular endo-thelial growth factor is a predictor of outcome in patients withacute myeloid leukemia. Blood 1999; 94: 3717–3721.

19 Presta M, Rusnati M, Belleri M, Morbidelli L, Ziche M, Ribatti D.Purine analogue 6-methylmercaptopurine riboside inhibits earlyand late phases of the angiogenesis process. Cancer Res 1999; 59:2417–2424.

20 Isacchi A, Statuto M, Chiesa R, Bergonzoni L, Rusnati M, Sarmi-entos P, Ragnotti G, Presta M. A six-amino acid deletion in basicfibroblast growth factor dissociates its mitogenic activity from itsplasminogen activator-inducing capacity. Proc Natl Acad Sci USA1991; 88: 2628–2632.

21 Liekens S, Neyts J, De Clercq E, Verbeken E, Ribatti D, PrestaM. Inhibition of fibroblast growth factor-2-induced vascular tumorformation by the acyclic nucleoside phosphonate cidofovir. Can-cer Res 2001; 61: 5057–5064.

22 Presta M, Maier JAM, Ragnotti G. The mitogenic signaling path-way but not the plasminogen activator-inducing pathway of basicfibroblast growth factor is mediated through protein kinase C infetal bovine aortic endothelial cells. J Cell Biol 1989; 109:1877–1884.

23 Montesano R, Pepper MS, Vassalli J-D, Orci L. Phorbol ester


1499induces cultured endothelial cells to invade a fibrin matrix in thepresence of fibrinolytic inhibitors. J Cell Physiol 1987; 132:509–516.

24 Pepper MS, Montesano R, Vassalli J-D, Orci L. Chondrocytesinhibit endothelial sprout formation in vitro: evidence for involve-ment of a transforming growth factor-beta. J Cell Physiol 1991;146: 170–179.

25 Ribatti D, Urbinati C, Nico B, Rusnati M, Roncali L, Presta M.Endogenous basic fibroblast growth factor in the vascularizationof the chick embryo chorioallantoic membrane. Dev Biol 1995;170: 39–49.

26 Ribatti D, Gualandris A, Bastaki M, Vacca A, Roncali L, Presta M.A new model for the study of angiogenesis and antiangiogenesisin the chick embryo chorioallantoic membrane (CAM): the gelatinsponge/CAM assay. J Vasc Res 1997; 34: 455–463.

27 Vacca A, Ribatti D, Iurlaro M, Albini A, Minischetti M, BussolinoF, Pellegrino A, Ria R, Rusnati M, Presta M, Vincenti V, PersicoMG, Dammacco F. Human lymphoblastoid cells produce extra-cellular matrix-degrading enzymes and induce endothelial cellproliferation, migration, morphogenesis, and angiogenesis. Int JClin Lab Res 1998; 28: 55–68.

28 Brooks PC, Clark RAF, Cheresh DA. Requirement of vascular inte-grin �v�3 for angiogenesis. Science 1994; 264: 569–571.

29 Bennet JM, Catovsky D, Daniel MT, Flandrin G, Galton DA, Gral-nick HR, Sultan C. Proposed revised criteria for the classificationof acute myeloid leukemia: a report of the French–American–Bri-tish Cooperative Group. Ann Intern Med 1985; 103: 620–625.

30 Kubota Y, Kleinman HK, Martin GR, Lawley TJ. Role of lamininand basement membrane in the morphological differentiation ofhuman endothelial cells into capillary-like structures. J Cell Biol1988; 107: 1589–1598.

31 Gualandris A, Rusnati M, Belleri M, Nelli EE, Bastaki M, Molinari-Tosatti MP, Bonardi F, Parolini S, Albini A, Morbidelli L, ZicheM, Corallini A, Possati L, Vacca A, Ribatti D, Presta M. Basicfibroblast growth factor expression in endothelial cells: anautocrine mechanism for angiogenesis and angioproliferative dis-eases. Cell Growth Differ 1996; 7: 147–160.

Leukemia

32 Folkman J. Tumor angiogenesis. Adv Cancer Res 1985; 43:175–203.

33 Marino MW, Dunbar JD, Wu L-W, Ngaiza JR, Han HM, Guo D,Matsushita M, Nairn AC, Zhang Y, Kolesnick R, Jaffe EA, DonnerDB. Inhibition of tumor necrosis factor signal transduction inendothelial cells by dimethylaminopurine. J Biol Chem 1996; 271:28624–28629.

34 Giannini G, Di Marcotullio L, Zazzeroni F, Alesse E, Zani M,T’Ang A, Sorrentino V, Screpanti I, Frati L, Gulino A. 2-Aminopur-ine unravels a role for pRB in the regulation of gene expression bytransforming growth factor � . J Biol Chem 1997; 272: 5313–5319.

35 Buchner T, Hiddemann W, Wormann B, Loffler H, Gassmann W,Haferlach T, Fonatsch C, Haase D, Schoch C, Hossfeld D, Leng-felder E, Aul C, Heyll A, Maschmeyer G, Ludwig WD, SauerlandMC, Heinecke A. Double induction strategy for acute myeloid leu-kemia: the effect of high dose cytarabine with mitoxantroneinstead of standard-dose cytarabine with daunorubicin and 6-thi-oguanine: a randomized trial by the German AML CooperativeGroup. Blood 1999; 93: 4116–4124.

36 Fiedler W, Graeven U, Ergun S, Verago S, Kilic N, StockschladerM, Hossfeld DK. Vascular endothelial growth factor, a possibleparacrine growth factor in human acute myeloid leukemia. Blood1997; 89: 1870–1875.

37 Bellamy WT, Richter L, Frutiger Y, Grogan TM. Expression of vas-cular endothelial growth factor and its receptors in hematopoieticmalignancies. Cancer Res 1999; 59: 728–733.

38 Rafii S, Shapiro F, Pettengell R, Ferris B, Nachman RL, Moore MA,Asch AS. Human bone marrow microvascular endothelial cellssupport long-term proliferation and differentiation of myeloid andmegakaryocytic progenitors. Blood 1995; 86: 3353–3363.

39 Roboz GJ, Dias S, Lam G, Lane WJ, Soignet SL Warrell RP Jr, RafiiS. Arsenic trioxide induces dose- and time-dependent apoptosisof endothelium and may exert an antileukemic effect via inhibitionof angiogenesis. Blood 2000; 96: 1525–1530.

40 Vacca A, Iurlaro M, Ribatti D, Minischetti M, Nico B, Ria R, Pelle-grino A, Dammaco F. Antiangiogenesis is produced by nontoxicdoses of vinblastine. Blood 1999; 94: 4143–4155.

anti-angiogenic activity of the purine analog 6-thioguanine

Documents